The fundamentals of knowledge operations
The Action-Knowledge-Information (AKI) model depicts the fundamental loop of ongoing sensing, sense-making, and response.
One of the most powerful models for thinking about knowledge is also one of the simplest. Known as the Action-Knowledge-Information (AKI) model, developed by David Williams, the model depicts the fundamental loop of ongoing sensing, sense-making, and response by all sentient systems:
As previously discussed, a system’s environment is everything that sits outside the system boundary, while the domain is the immediate portion of the environment that can exert detectable system effects.
These allow us to concisely describe the foundations of the AKI model:
Action — The activities of a system intended to affect its domain
Information — A received representation of the system’s environment
Knowledge — An integrated capability to choose between and execute meaningful actions
We can further enrich the AKI model by introducing the idea of a proxy, an agent that reliably acts, or receives information, on our behalf. For example, the Bureau of Meteorology interprets and shares information about the weather with us rather than us observing the weather ourselves. From this, we can then introduce the categories of direct and mediated acts and interactions:
Direct information is information received from a system’s own experiences, allowing it to construct its own representation of those experiences
Mediated information is information that is first processed by a proxy before being passed on to a system
Direct knowledge is knowledge derived from the immediate feedback loop of a system’s experience in its environment
Mediated knowledge is knowledge that is developed from the mediated information provided by a proxy
Direct action is action that is performed by a system itself
Mediated action is where a system instructs or enforces that an action be carried out by a proxy
The distinction is important because mediated action, knowledge, and information are inherently representational: they do not act on or receive signals from reality, but an interpretation of reality that is abstracted and defined by the proxy.
The move from direct to mediated interactions may seem a poor trade-off: surely it is better to work with the fire-hose of unfiltered reality than an impoverished representation using symbols and signals? But there are several critical advantages of mediated interactions:
They are not time bound — Impacts of received information or actions taken can happen minutes, days, or years after the triggering event. This was the revolutionary concept embodied by writing systems.
They are scalable — Impacts beyond the analysed system can occur at a scale far beyond its direct power as the ability to replicate increases.
They promote alignment — Any representation must necessarily simplify the infinite complexity of reality into a subset of those options to communicate. Over time, greater consensus arises on how to interpret communications, in turn supporting common interpretations of reality across systems, which expand opportunities for coordination and partnership.
While direct information, knowledge, and action remain essential for truth-seeking, most larger systems rely heavily upon mediated interactions to expand their reach, persistence, and alignment. Not coincidentally, these are the very attributes which have enabled organisations and societies to achieve outcomes far beyond that possible from any single agent’s AKI loop.
Paradoxically, our reliance upon proxies has now become a vulnerability, ever since the cost of detecting impersonated or fake interactions with a proxy has significantly increased compared to the cost of detection (as evidenced by the number of participants fooled by the many high-quality videos now possible with AI).
If it becomes impossible to rely upon proxies, systems will instead have to massively scale up their own capabilities in direct interactions, regardless of how inefficient it may be. Re-establishing trust at scale is shaping up to be one of the key challenges for the 21st century. Without an accurate and adaptive method for establishing and maintaining proxy interactions, organisations as we know them may cease to be.